bart-base-asqa-ob

This model is a fine-tuned version of facebook/bart-base on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 1.8291
  • Rougelsum: 13.0645

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

  • learning_rate: 5e-06
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 20
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Rougelsum
No log 1.0 355 1.9076 13.1069
2.2336 2.0 710 1.8749 13.0551
2.048 3.0 1065 1.8580 13.1287
2.048 4.0 1420 1.8413 13.1473
2.0003 5.0 1775 1.8451 13.1264
1.9423 6.0 2130 1.8360 13.0959
1.9423 7.0 2485 1.8372 13.1289
1.8894 8.0 2840 1.8275 13.1359
1.8568 9.0 3195 1.8241 13.0983
1.8279 10.0 3550 1.8279 13.0184
1.8279 11.0 3905 1.8275 13.1177
1.7871 12.0 4260 1.8279 13.0871
1.7666 13.0 4615 1.8295 13.0992
1.7666 14.0 4970 1.8291 13.0645

Framework versions

  • Transformers 4.23.0.dev0
  • Pytorch 1.12.1+cu102
  • Datasets 2.4.0
  • Tokenizers 0.12.1
Downloads last month
46
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support